The Radio Galaxy Zoo Data Release 1: classifications for 75,589 sources

Transcription

1 The Radio Galaxy Zoo Data Release 1: classifications for 75,589 sources O. Ivy Wong & Radio Galaxy Zoo Team ICRAR/University of Western Australia SPARCS VII the precursors awaken, 19 July

2 All-sky below declinations +20 deg Expect 70 million radio sources Norris Survey Area riding on the EMU's back... Sensitivity limit (mjy) 2

3 Motivation There is nothing quite as useless as a radio source. Condon, 2013 Translation: to understand how galaxies grow supermassive black holes & evolve, one needs context from multiwavelength observations 3

4 How to match 70 million radio sources to their hosts? humans (astronomers/their students) software matching algorithms - current matching algorithms work for 90% of sources (Norris'12) so what about the other 7 million sources? advance machine learning algorithms more humans? 4

5 Path ahead... Clear need for new automated methods to make accurate cross ids But, there exists many exotic radio morphologies that are not well catalogued/documented Step 1: create a large dataset with different radio source morphologies 5

6 radio.galaxyzoo.org 6

7 Combining archival datasets + Cutri Becker, White & Helfand Franzen+ 2015, Norris+2006 Lonsdale

8 Citizen scientists (radio.galaxyzoo.org) 8

9 radio.galaxyzoo.org 1) Examine radio & IR images 2) Identify radio source components 3) Mark location of host galaxy stay tuned for Julie's talk 9

10 Radio Galaxy Zoo Data Release 1 Classifications between Dec 2013 & March ,214 registered citizen scientists >1.69 million classifications 75, 589 source classifications 74,627 from FIRST 962 from ATLAS Catalogue of radio source components + WISE host for all sources with consensus 0.65 Wong in prep 10

11 Comparing FIRST & ATLAS Median consensus of FIRST much higher than those of ATLAS Why? Match resolution and confusion matters! Wong in prep 11

12 Checking DR1 Visual verification of 1001 sources from * Big thanks to Larry Rudnick, HeinzAndernach, Stas Shabala, Ray Norris & Jesse Swan for doing this with me FIRST and ~2123 sources from ATLAS* DR1 FIRST consensus levels correlate with the average fraction of classifications verified (within limits of obs) The 6 of us can disagree at a 5 10% level However, ATLAS data shows that this verified fraction is misleading because confusion cross id degeneracies Wong in prep; 12

13 Quantified reliability of DR1-FIRST The verified fraction (f) cannot quantify Re(single parameter) Re(single parameter) = mean f, weighted by n_sources per bin DR1 cat = min reliability of ~44% DR1 w/o WISE W1 but bright W1 also an issue Wong in prep 13

14 Technical lessons What increases confusion/degeneracies in citizen scientists' classifications? Radio IR resolution difference ATLAS, FIRST: 12.2'' x 7.6'', 5'' SWIRE, WISE(W1) : 1.2'', 6'' Deep observations (radio & IR) ATLAS, FIRST: 1 = 13, 150 Jy/beam SWIRE, WISE(W1): 5 = 7.3, 80 Jy Wong in prep 14

15 Science with Radio Galaxy Zoo Data Release 1 Wong in prep 15

16 6,202 DR1 sources with good W1,W2,W3 Comparable to LARGESS Jarrett Wong in prep Note. DR1 ~6.3 times larger than LARGESS 16

17 1.4 GHz vs 12 microns Dividing 1 component DR1 sample using WISE colours Grey contours: QSO Grey contours: QSO Orange: Starburst/SF Orange: Ellipticals Grey contours: QSO Orange: Interm disks SF arm Wong in prep 17

18 Radio-quiet radio silent Consistent 1.4 GHz vs 12 micron results from the MIXR sample AGN Ellipticals Star formers Spirals/ intermediate disks Mingo+ 2016; see also Wong Note. DR1 ~3.9 times larger than MIXR 18

19 IR properties in terms of radio components 1 component 2 components 3 components QSO WISE colours dominate n_comp = 1 3, relative QSO fraction = 53% 83% Relative fraction of Spiral/starburst host decrease from 36% 11% Wong in prep 19

20 Radio source projected angular size 1,015 DR1 sources with more than 1 radio component Median projected angular size = 43.1'' Wong in prep 20

21 IR properties of small & large sources Small (< 43.1'') Large (> 43.1'') Small radio sources associated mostly with QSO WISE colours More elliptical hosts for radio sources with large angular sizes Wong in prep 21

22 Summary Radio Galaxy Zoo Data Release 1 >75,000 source classifications with a minimum reliability of ~44% w/o IR host, 6202 (low z) sources have good W1, W2, W3 Many sources w spiral/starburst IR colours can have significant radio emission from a radio quiet AGN Multi component sources: more likely to have QSO IR colours. BUT, elliptical fraction for sources w larger angular sizes Wong in prep 22